The research work of this thesis is motivated by a desire to render views of 3D environments in real-time on mobile devices. Considering the weak support for 3D hardware accelerations of mobile devices, it is hard to render geometry-based models at an interactive rate. The utilization of image-based rendering techniques in mobile devices is presented in this thesis. In order to have a systematical knowledge of image-based rendering, we briefly review different image-based rendering techniques. Although image-based rendering is advantageous to geometry-based rendering in certain circumstances, further tailoring is still required to immigrate it to mobile devices. This thesis presents a technique for running image-based rendering tasks on mobile devices. This technique comprises a lightweight rendering algorithm and a local cache management mechanism. Results have shown that the proposed algorithm requires less CPU cycles compared to other algorithms, and that the local buffer management mechanism is robust to communication errors.

A high temporal resolution pollen diagram from a lake in the middle-Arctic region of the Boothia Peninsula, Nunavut, Canada, documents the history of the regional vegetation and climate for the past 7200 years. A diatom sequence had been previously prepared from this core. Major tundra pollen taxa in the core include Cyperaceae and Salix, with Cyperaceae comprising over 50% of the pollen in the early and late Holocene. Tree pollen, transported from far to the south, comprised a large percentage of the pollen sum, with Pinus accounting for 30% of the pollen in some levels of the core. Pollen percentages and concentrations of taxa typical of the middle-Arctic were highest in the mid-Holocene, corresponding to warm conditions. Decreasing pollen concentrations indicate cooling temperatures, with more rapid decreases occurring around 4200, 3800-3400, and 2500 cal yr BP. Pollen percentages of Salix, Cyperaceae, and Artemisia increased in the past 35 years in response to global warming. Reconstructions of July temperature using the modern analog technique showed the mid-Holocene (5800-2800 cal yr BP) was approximately 1°C higher than during the past 1000 years.

Between 1977 and 2007, the Kaskawulsh Glacier underwent an overall decrease in volume of between 2.76 km3 water equivalent (we) and 4.60 km3 we, and a decrease in area of 2.27%. Volume losses are consistent with changes observed at most glaciers in the Yukon/Alaska region which are thinning and/or retreating (Arendt et al. 2006; Chen et al. 2006a; Larsen et al. 2007; Molnia 2007). The rate of volume change over the periods 1977-1995 and 1995-2007 remained constant at -0.52 km3 yr-1 we, while between 2000 and 2007 the glacier gained volume at a rate of 1.04 km3 yr-1 we. Gains in the recent 2000-2007 period result from prominent thickening in the accumulation area above 1989 m (+16.9 m) and minor thickening in the ablation area (+0.5 m). The observed thickening pattern is similar to patterns observed in Greenland (Chen et al. 2006b; Johannessen et al. 2005; Luthcke et al. 2008b), Antarctica (Davis et al. 2005), the Karakoram Himalaya (Hewitt 2005), New Zealand (Chinn 1999; Hooker and Fitzharris 1999), Scandinavia (Bamber et al. 2004; Chinn et al. 2005, Nesje et al. 2000), and Alaska (Arendt et al. 2008; Muskett et al. 2(03). Between 1956 and 2007, the glacier terminus retreated by an average of 655 m (13 m yr-1). No scaling ratio between terminus retreat and volume change could be established for the Kaskawulsh Glacier for the period 1977-2007.

The links between climate and permafrost are well known, but the precise nature of the relationship between air and ground temperatures remains poorly understood, particularly in complex mountain environments. Although previous studies indicate that elevation and potential incoming solar radiation (PISR) are the two leading factors contributing to the existence of permafrost at a given location, additional factors may also contribute significantly to the existence of mountain permafrost, including vegetation cover, snow accumulation and the degree to which individual mountain landscapes are prone to air temperature inversions. Current mountain permafrost models consider only elevation and aspect, and have not been able to deal with inversion effects in a systematic fashion. This thesis explores the relationship between air and ground surface temperatures and the presence of surface-based inversions at 27 sites within the Wolf Creek basin and surrounding area between 2001 and 2006, as a first step in developing an improved permafrost distribution TTOP model. The TTOP model describes the relationship between the mean annual air temperature and the temperature at the top of permafrost in terms of the surface and thermal offsets (Smith and Riseborough, 2002). Key components of this model are n-factors which relate air and ground climate by establishing the ratio between air and surface freezing (winter) and thawing (summer) degree-days, thus summarizing the surface energy balance on a seasonal basis. Here we examine (1) surface offsets and (2) freezing and thawing n-factor variability at a number of sites through altitudinal treeline in the southern Yukon. Thawing n-factors (nt) measured at individual sites remained relatively constant from one year to the next and may be related to land cover. During the winter, the insulating effect of a thick snow cover results in higher surface temperatures, while thin snow cover results in low surface temperatures more closely related to the winter air temperatures. The application of n-factor modeling techniques within the permafrost region, and the verification of these techniques for a range of natural surfaces, is essential to the determination of the thermal and physical response to potential climate warming in permafrost regions. The presence of temperature inversions presents a unique challenge to permafrost probability mapping in mountainous terrain. While elsewhere the existence of permafrost can be linearly related to elevation, the presence of frequent inversions challenges this assumption, affecting permafrost distribution in ways that the current modeling techniques cannot accurately predict. At sites across the Yukon, inversion-prone sites were predominantly situated in U-shaped valleys, although open slopes, mid-slope ridges and plains were also identified. Within the Wolf Creek basin and surrounding area, inversion episodes have a measurable effect on local air temperatures, occurring during the fall and winter seasons along the Mount Sima trail, and year-round in the palsa valley. Within the discontinuous permafrost zone, where average surface temperatures are often close to zero, even a relatively small change in temperature in the context of future climate change could have a widespread impact on permafrost distribution.

Speleothems are rare in high-latitude and high-altitude caves, which is why Bear Cave in the north-western Yukon Territory is particularly unique, as it houses some of the oldest and highest latitude speleothem in the world. In this study, a detailed petrographic and geochemical study was conducted along the profile of a 68-cm long late-Miocene flowstone from Bear Cave (BC1) to reveal the processes that took place at the time of deposition, in addition to those that followed, in aims to determine its paleoclimatic suitability. These studies suggest that softer facies are generally representative of disequilibrium conditions, where in-filled textures provide evidence for diagenic phenomena; these processes ultimately obscure the original climate signal, compromising the integrity of the flowstone in terms of its paleoclimatic suitability. Conversely, harder facies are likely deposited in isotopic equilibrium and resistant to post-depositional diagenesis, and are therefore more reliable for detailed paleoclimatic analysis. The variability of the calcite sequences in the profile of BC1 imply that environmental conditions have been considerably variable of the course of deposition, reflecting alternating cool/dry (softer facies) and warm/wet (harder facies) climatic conditions. Results from radiogenic 4He-dating constrain the timeframe of deposition, with ages centering around 9.35 +/- 0.52 Ma, which are in line with regional geomorphic interpretations. This study emphasizes the importance of a complimentary petrographic study in speleothem geochemical studies used in paleoclimatic and paleoenvironmental research.

Variations in the Earth's climate occur on many time and space scales. A recent focus of paleoclimate research is the so-called 1500-year North Atlantic quasi-periodic cycle, and has revolved around three main themes. First, what are the underlying causes and physical mechanisms governing these millennial-scale variations? Next, are they global or restricted to certain sensitive regions of the planet? Last, what is the magnitude of the temperature changes of these variations, and do they vary in time and space? This dissertation explores millennial-scale climate variability in North America during the past 14,000 years using a dense network of fossil pollen data, which is used as proxy for climate variations. Three independent approaches are used to quantify these changes. A mixture modelling analysis of radiocarbon dates on pollen transitions, a principal component analysis of pollen diagrams from all of North America, and a mean July temperature reconstruction based on the method of modern analogue (MAT) all reveal millennial-scale climate variability throughout North America during the past 14,000 years. The identified transitions generally correlate well with other proxy-climate records from the North Atlantic region. However, certain mismatches occurred particularly at 9, 6 and 4 ka BP. If we assume the dominant millennial-scale period is 1150-years, the records become more consistent. North American temperature variability was not unidirectional nor uniformly distributed in space, suggesting large-scale ocean-atmospheric reorganizations at the transitions. Correlation between the proxy-climate and cosmogenic nuclide records supports a variable solar output hypothesis as the fundamental cause for century to millennial-scale climate variability. The mean July temperature of North America varied on the order of 0.2 to 0.4°C during the Holocene and 0.4° and 0.6°C during the deglaciation. Temperature was more variable during the late glacial, possibly due to the impact on the climate of massive meltwater pulses into the North Atlantic, further amplified through ocean dynamic processes. Recent global warming estimated as an increase in temperature of 0.4--0.6°C, is therefore greater than the estimated natural variation of the past 10,000 years, providing further evidence that recent increases in atmospheric CO2 have played a major role in modern warming.

In most cases contaminants are concealed in soil and under vegetation and therefore can not be measured directly by remote sensing. However, soil contaminants were detected using the spectral red-edge to indicate vegetation stress caused by the presence of the contaminants. An improved red-edge position (REP) was developed and gave a slight improvement in the predictive capability over existing indices and an effective additional diagnostic indicator of soil contamination was found to be the spatial pattern of the REP. Where an area had high levels of hydrocarbon in the soil it also had a high level of variation. The indication was that spatial variation of spectral indices (especially the REP) may be more useful than the spectral index value for the detection and mapping of soil contamination. Field analysis and radiative transfer modelling (using a coupled leaf and canopy model, LIBSAIL) showed the influence of vertical layering in the grassland canopy. The influence of a vegetated under-storey on the red-edge was found to be greatest when different absorption spectra were present and high within-the-leaf scattering. The former defined wavelength positions of features while the later determined if they were resolvable in a spectrum. This greater understanding of the grassland canopy identified the importance of fully surveying vegetation canopy structure, especially in complex, multi-layered canopies such as those found with contamination. With this understanding of what the red-edge can reveal, remote sensing is an effective tool for the detection of contamination.

The recent increase in large fires in southwestern forests has prompted concern regarding their ecological consequences. Recognizing the importance of spatial patterns in influencing successional processes, I asked: (1) How do large fires change plant communities?; (2) What are the implications of these changes for ponderosa pine forests?; and (3) What is the relationship of fire severity to gradients of climate, fuels, and topography? To address the first two questions, I studied succession in the woody plant community at two sites that burned in high-severity fire: La Mesa fire in northern New Mexico (1977) and Saddle Mountain in northern Arizona (1960). After large fires, abiotic conditions, associated prefire plant distributions, and spatial patterns of burning interacted to result in particular successional outcomes. Variation in abundance and diversity of species that spread from a refuge of seed sources remaining after the fire followed the model of wave-form succession. I investigated the implications of large fires for ponderosa pine by examining the influence of spatial patterns of burning on regeneration. Tree density corresponded most closely with particular scales of seed dispersal kernel and neighborhood severity metrics. Spatial patterns of burning remained influential even after consideration of variables describing subsequent burning and the physical and biotic environment. Age structure of young forests indicated that populations spread in a moving front and by long-distance dispersal. To explore the relationship between fire severity and climate, I investigated how the spatial heterogeneity of high-severity patches varied among 20 fires across gradients in fire size and climate. The largest fires generally occurred during cool dry La Niña climates, however, several fires deviated from this trend. Some spatial properties of severity did not correspond to fire size or to changes in climate. Characteristics of fuels and topography altered spatial patterns of severity, but interactions with extreme burning conditions may have disrupted these local influences in both La Niña and El Niño fires. Spatial patterns of fire severity are central to understanding ecological dynamics following large fires in southwestern forests. Moreover, simplistic assumptions regarding the relation of fire severity to fire size and climate should be viewed with caution.

Paleoglaciological reconstructions based on glacial geological and geomorphological traces are used to test and constrain numerical models of ice sheet extent and dynamics. MAGIC-DML (“Mapping, Measuring and Modelling Antarctic Geomorphology and Ice Change in Dronning Maud Land”) project is trying to reconstruct the timing and pattern of ice surface elevation changes since the mid-Pliocene across western Dronning Maud Land, East Antarctica. The study area has sparse pre-existing field data and considerable ice sheet model uncertainties. A remote sensing-based mapping of glacial geomorphology on nunataks and structures on the ice sheet surface is presented for a coastal-inland transect including Ahlmannryggen, Borgmassivet, and Kirwanveggen using high-resolution WorldView imagery. The primary aim of the study is to map traces of a thicker ice sheet on nunatak slopes that were formerly partly or entirely covered during ice surface highstands. Panchromatic and multispectral images were analysed in a multi-step procedure using ArcGIS, including image processing and mosaicking, visual feature recognition, and mapping. The identification of key landforms (such as till veneers and erratic boulders) required the adoption of some assumptions to differentiate, for example, till from regolith. Where patterned ground was mapped, we infer a presence of till rather than regolith because subglacial erosion is more likely to produce finer material than subaerial weathering. Very large boulders on plateau surfaces are mapped as erratics because they could not have been delivered by slope processes to local highpoints. However, the reliability of derived paleo-ice sheet reconstructions is limited by both the necessary assumptions and the absence of crosscutting relationships between landforms. At face value, the presence of till cover and erratics above the present ice surface on some nunataks indicate thicker ice in the past. According to the geomorphological mapping of the transect, in Kirwanveggen the former ice elevation was at least 100 m higher, in Borgmassivet the ice lowered more than 600 m and in Ahlmannryggen the ice was at least 300 m thicker. Additional mapping of structures on the ice sheet surface is used to yield target field routes for upcoming field season(s) to potential cosmogenic nuclide (CN) sampling locations. The chronology derived from CN dating will permit the delineation of ice sheet surface elevations as targets for ice sheet modeling. / MAGIC DML

Paleoglaciological reconstructions based on glacial geological and geomorphological evidence areused to constrain and test numerical models of ice sheet extent and dynamics. The MAGIC-DMLresearch project (“Mapping, Measuring and Modelling Antarctic Geomorphology and Ice Change, inDronning Maud Land”) is trying to reconstruct the timing and pattern of ice surface elevation changessince the mid-Pliocene across western Dronning Maud Land, East Antarctica. This reconstructionwill work as the basis for testing and constraining ice sheet numerical models to improve climateunderstanding in Antarctica. This master thesis project contributes to MAGIC-DML by adopting a high-resolution remote sensingbasedmapping of glacial geomorphology and ice sheet surface structures, for a coast-inland transectincluding the Ahlmannryggen, Borgmassivet, and Kirwanveggen nunatak ranges. The primary aimof this study is to investigate the glaciology and paleoglaciology of the study area, in order to mapevidence for a former thicker ice sheet on nunatak slopes and plateaus, and patterns of ice flow of thecurrent ice sheet surface. Meso-scale glacial landforms and ice flow features were identified andmapped using different remote sensing data sets: the LANDSAT Image Mosaic of Antarctica(LIMA), DigitalGlobe Worldview-2 (WV02) and Worldview-3 (WV03) panchromatic andmultispectral images, the Radarsat Antarctica Mapping Project (RAMP) Ice Surface Digital ElevationModel (DEM) version 2, and the Bedmap2 datasets. The satellite imagery was analysed in a multistepprocedure using ArcGIS, including image processing and mosaicking, visual feature recognition,and mapping. The identification of some key landforms required the adoption of assumptions, forexample in order to distinguish till cover from regolith or boulders derived from rock fall from glacialerratics. Present-day ice flow directions were traced according to the distribution of ice surfacefeatures such as blue ice areas, crevasse fields, longitudinal surface structures, and supraglacialmoraines. The occurrence of till cover and erratics above the present-day ice surface on somenunataks slopes and plateaus was considered indicative of a thicker ice sheet in the past. Paleo-iceflow directions were inferred from the proximity of locations to the closest ice streams, since thatlatter have been active since the Oligocene. Geomorphological and ice flow direction maps were obtained and used to infer the paleoglaciologyof the three nunatak ranges. Ice sheet thinning reconstructions reveal a minimum ice surface loweringof ~400–500 m in the Ahlmannryggen and Borgmassivet nunatak ranges, of ~300 m north of theKirwanveggen escarpment and of ~100 m on the edge of Amundsenisen polar plateau. The paleo-icesheet flow pattern probably differed from today, because ice flow has locally been influenced by anincreased topographical complexity, due to the thinning of the ice sheet and the emerging of nunatakoutcrops. According to dating studies conducted elsewhere in DML, the inferred ice surface decreasewas probably initiated in the Late Pliocene/Early Pleistocene, and continued after the Last GlacialMaximum interruption across the coastal sector of the ice sheet. The reliability of derived paleo-icesheet reconstructions, based on the mapping and interpretation of landforms, needs to be verified infuture field studies. This master thesis project has identified 34 well-suited locations for the samplingof erratic boulders and bedrock surfaces for cosmogenic nuclide (CN) surface exposure dating duringthe MAGIC-DML 2017/18 field season. The chronology derived from CN dating and fieldverification of the presented mapping will permit the delineation of ice sheet surface elevations astargets for ice sheet modelling. / MAGIC DML